Intelligent Control System Of Granary Environment Based On Grain Condition

Grain is the basic condition to ensure the survival of Chinese citizens.With the development of science and technology in grain production,China's grain output is increasing.However,in the process of grain storage,the storage temperature,humidity and pests cause huge losses of grain,so timely access to the environmental conditions for grain storage is the basis of "scientific grain preservation".Traditional grain situation monitoring system collects temperature and humidity information.However,in the case of pest detection,the sampler method was usually used and then the species identification was carried out by skilled person.The system lacks the function of automatic image acquisition and recognition for stored grain pests.It also lacks the function of automatic control when the granary environment is abnormal.Based on the above problems,our system integrated the functions of intelligent control,image acquisition and recognition of stored grain pests on the basis of collecting grain situation.The system can automatically control the operation of mechanical ventilation equipment according to the acquired temperature and humidity data and identify the pest species according to the collected insect images.The main research contents and work are as follows:(1)A hardware system for grain information acquisition and a human-computer interaction software were designed.By using MCU,temperature and humidity sensor and image acquisition device,the system obtains the temperature and humidity data of each node of the granary as well as the images of stored grain pests.The grain situation information acquisition system transmits the data to the PC software through TCP/IP protocol and realizes the collection and visualization of grain situation information.(2)According to the different environment required by different grain storage,the human-computer interaction software was added to customize the temperature and humidity threshold of the current granary.The set threshold value was compared with the grain condition data collected by the system and then the corresponding control instructions were issued to realize the automatic operation of mechanical ventilation equipment,so as to realize the function of intelligent regulation of the granary environment according to the current grain condition.(3)The collected images of stored grain pests were uploaded to the cloud server to realize the recognition of stored grain pests.The cloud server contains the deep learning algorithm proposed for the common images of six kinds of stored grain pests.By introducing diliated convolution and short connections into the network,two types of blocks were designed.The number of parameters and calculation cost of the network were not increased while the receptive field area was expanded.Re LU was used as the activation function and Adam optimizer and Softmax classifier.By introducing Dropout,overfitting phenomenon of the model was reduced.To ensure the objective and fair results of the experiment,the model was verified by K-Fold cross validation.After 5-Fold cross validation,the average value of ACC was 96.72%,TPR was 90.17%,TNR was 98.03%,F1 score was 90.17% and AUC was 97% on the original datasets.Experimental results also demonstrate that compared with other recognition methods for the same datasets,the proposed method has the highest accuracy.